Innovative developments in systems condition monitoring

It is shown in this paper that classical approach to condition monitoring of critical systems can be supplemented by holistic models, which enable the best symptom of condition to be chosen and the evolution of the sytem condition to be simulated. When this is not possible, and where large symptom data base exist, we can apply singular value decomposition (SVD) as the newest data mining procedure to obtain symptom evolution model. By using SVD it is possible to have two additional independent fault discriminant: SD and SG, with the high dynamics of evolution, Moreover we can add life count as the first approximation of a logistic vector describing the unit life history. It is also possible to use the value of pseudo - determinant of a symptom observation matrix, and to correlate between this new discriminant and the symptom observation matrix to minimise the redundancy of symptom measuring space.